Ist., I, pfig 252.

Changes in the levels of indigenous microbial association of lettuce and cabbage salads were monitored at 8oC. Pseudomonas spp. (4.8-5.5 log CFU/g) and Enterobacteriaceae (4.9-5.8 log CFU/g) comprised the microbial flora of both salads (Figure 4-2). Yeasts and lactic acid bacteria were also present but at the lower levels of <2.3-2.6 and <1.8-2.0 log CFU/g, respectively (results not shown). Following treatment of vegetables with 200 ppm sodium hypochlorite, the reductions of pseudomonads ranged from 1.8

Figure 4-2 Populations (log CFU/g) of the indigenous microflora of lettuce (top) and cabbage (bottom) before and after treatment with 200 ppm sodium hypochlorite for 5 min and washing with sterile water for 15 min.

to 2.2 log CFU/g and that of Enterobacteriaceae from 2.9 to 3.8 log CFU/g (Figure 4-2). Both products were organoleptically unacceptable after 10-12 days, when the specific spoilage microorganisms (pseudomonads) had reached their maximum level of 9.4 ± 0.2 or 8.1 ± 0.3 log CFU/g in lettuce and cabbage salad, respectively (Figure 4-3 and Figure 4-4).

The microbial load of minimally processed salads usually consists of microorganisms related to soil or humans, such as pseudomonads, enterobacteria, lactic acid bacteria and yeasts (Randazzo, et al., 2009; Abadias, et al., 2008). Although disinfection of fresh produce aims to reduce

Figure 4-3 Growth of 1000 (i) or 1-4 cells (ii) of Listeria monocytogenes per sample of lettuce (A) or cabbage (B) salad at 8°C. () 1000 cells of L.

monocytogenes, () total viable counts, () pseudomonads, () Enterobacteriaceae, () 1-4 cells of L. monocytogenes. Counts of the indigenous microflora (A-i, B-i) also refer to (A-ii, B-ii). The initial levels of L.

monocytogenes in samples inoculated with 1-4 cells (-1 to -0.4 log CFU/g;

graph A-ii, B-ii), are rounded to 0 log CFU/g, with detection limit of 0.7 log CFU/g.

the epiphytic flora and prolong the shelf-life of the final products (Delaquis, et al., 2004; Koseki and Itoh, 2001), reducing the epiphytes may also weaken their antagonistic activity against pathogens that survive the intervention or potentially contaminate the products during further processing.

With regards to the growth of pathogens, L. monocytogenes cells were able to grow on both salads, regardless of the inoculum level. Increase of L.

0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 L og CF U/g Time (days) 0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 L og CF U/g Time (days) 0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 L og CF U/g Time (days) 0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 L og CF U/g Time (days) A-i A-ii B-i B-ii

1000 cells per sample 1-4 cells per sample

L ettu ce C ab b ag e

Figure 4-4 Growth of 1000 (i) or 1-4 cells (ii) of Salmonella Typhimurium on lettuce (A) and cabbage (B) salad at 8°C. () 1000 cells of S. Typhimurium, () total viable counts, () Pseudomonadaceae, () Enterobacteriaceae, () 1-4 cells of S. Typhimurium. Counts of the indigenous microflora (A-I, B-i) also refer to (A-ii, B-ii). The initial levels of S. Typhimurium in samples inoculated with a 1-4 cells (-1 to -0.4 log CFU/g; graph A-ii, B-ii), are rounded to 0 log CFU/g.

monocytogenes starting from 1000 cells/sample occurred with limited variation (standard deviation <0.5 log CFU/g), while growth initiating from a few (1-4) cells/sample ranged from <0.7 log CFU/g (detection limit) to 3.4 log CFU/g (Figure 4-3). Similar observations were made on the growth variability of the high and low inocula of S. Typhimurium on lettuce, whereas no growth of this microorganism was obtained on cabbage ( Figure 4-4). It is notable however, that the total log-increase (LI) from 1-4 cells/sample of both pathogens was higher compared with that of 1000 cells/sample. More

0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 Log C FU/g Time (days) 0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 Log C FU/g Time (days) 0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 Log C FU/g Time (days) 0.0 2.0 4.0 6.0 8.0 10.0 0 2 4 6 8 10 12 14 Log C FU/g Time (days)

1000 cells per sample 1-4 cells per sample

L ettu ce C ab b ag e

1000 cells per sample 1-4 cells per sample

A-i A-ii

3.4 log CFU/g on lettuce and cabbage salad, respectively, while the corresponding increase from 1000 cells was LI (t=12 days) = 2.1 and 1.8 log

CFU/g (Figure 4-3). Similarly, 1-4 cells/sample of S. Typhimurium showed LI

(t=12 days) = 3.0 log CFU/g on lettuce salad, whereas only LI (t=12 days) = 1.1 log

CFU/g was observed when starting from 1000 cells/sample ( Figure 4-4). In all cases, the growth of pathogens ceased when the epiphytic microflora reached its maximum population density.

A typical fresh-cut salad sample consists of various micro-environments with different micro-architectures such as liquid phase (juices), solid phase (tissue) or combination of those. The diffusion, the variety and the intensity of potentially stressful factors among these micro-environments may vary significantly (Wilson et al., 2002; Leveau and Lindow, 2001), along with the physiology of individual cells (Dupont and Augustin, 2009; Koutsoumanis, 2008). During random localization of individual cells on vegetable samples, the probability of two cells, which have been inoculated on two independent samples, to occur in a similar micro-environment is decreased, and in combination with their individuality, significantly different growth potential for each cell could be expected. Conversely, distributing higher inocula on different spots of independent samples increases the similarities in spatiotemporal occurrence of cells at optimum (i.e., highest potential to grow in adverse conditions) or similar physiological state. Thus, the growth variability observed between such samples is lower.

An important observation with regards to food safety was the higher LI that a few cells of the pathogens exhibited during storage on both vegetables, compared to that of the higher initial inoculum (i.e., 1000 cells/sample). For instance, the percentage of lettuce or cabbage samples, inoculated with 1-4 cells/sample of L. monocytogenes and exceeding the microbiological criterion of 100 CFU/g, varied from 5 to 70%, depending on storage time. Conversely, the inoculum of 1000 cells/sample showed total LI equal or lower than this criterion (i.e., ≤ 2 log CFU/g). A critical factor affecting the capacity for growth of various initial pathogen levels on a food surface is the interaction with the

epiphytic flora. The observation that the growth of L. monocytogenes and S. Typhimurium ceased when the background flora of salads reached the maximum population density (Figure 4-3and Figure 4-4), may be associated with the so-called ―Jameson effect‖ (Mellefont, et al., 2008), due to the competition for nutrients, or the production of microbial metabolites by the indigenous flora. In contrast, as detailed in section 4.3.2, no such inhibition of the pathogens was observed in the sterile extracts, apparently due to the absence of the background flora. However, given that the epiphytes reached their maximum population density at the same storage time in all batches of each vegetable, (i.e., 10 and 8 days for lettuce and cabbage, respectively), the growth of the pathogens ceased simultaneously, regardless of the initial inoculation level. Thus, the differences in total log increase between samples inoculated with high and those with low inocula could be attributed to differences in both the growth rate and/or lag time. According to previous reports, the growth rate is considered independent of the initial inoculum size (Robinson et al., 2001). Therefore, the aforementioned differences in LI of different inocula should only be due to the variability in individual lag times. It needs to be stressed however that the majority of studies on the effect of the inoculum size on the growth kinetics of pathogens has been performed on liquid media, where after each bacterial division, the daughter cells drift away from the mother cell. Thus, the possibility of ―crowding effect‖, i.e., the growth deceleration due to interactions of closely located cells is limited, and this may explain why the growth rate of planktonic cultures is not affected by their initial inoculum size. On the contrary, when cells are immobilized on a solid matrix, as is the case of fresh-cut salads, cells grow as colonies and hence, interactions may occur between cells within a colony (especially among cells in the outer and inner part of colonies) or between adjacent colonies. Such interactions are thought to increase with the proximity of colonies, i.e., with the population density, because the higher the population the lower the distance between colonies. This most likely explains the slower growth rate of 1000 cells/sample compared to that of a few cells/sample. Likewise, Thomas et al. (1997) reported that increasing the distance between adjacent

colonies of L. monocytogenes from 100 κm to 3000 κm, increased the cell density in the colonies from approximately 1.5 to 7 log CFU per colony, respectively. The above results imply that if log increase of pathogens on a food was judged only by challenge tests with unrealistically high inocula, as the worst case scenario, the actual risk would have been underestimated. Overall, although the risk posed by a few L. monocytogenes or S. Typhimurium cells on vegetable salads may be characterized by high variability, such assessments are more realistic and representative of a low contamination scenario. Considering also that the growth kinetics of the pathogens may differ significantly, depending on the initial inoculation level, challenge tests initiated from low number of cells (<5-10 cells) should be considered as a more reliable method to evaluate the real risk.